%輸入色譜數(shù)據(jù)P和對應(yīng)的目標(biāo)期望值T
b=[ 0.5709    0.0439    0.4996    0.2377    2.3461    0.0423
    1.0496    0.3980    1.6184    0.1573    0.9451    1.5614
    0.6880    0.2292    1.1794    1.1895    0.5214    0.2384
    0.6540    0.2017    1.6566    0.5587    0.4112    1.2298
    2.0277    0.3553    1.8477    1.1107    0.1547    0.2531
    1.5818    0.4356    1.6425    1.2554    0.2268    0.5861
    0.5384    0.4898    0.4886    0.4446    0.1228    0.8452
    0.1734    0.1885    0.0966    0.1141    0.0819    0.2619
    0.1394    0.3331    0.0812    0.0585    0.0935    0.3528
    0.3665    0.1494    0.3549    0.2927    0.1035    0.1513
    0.1424    0.2386    0.1759    0.1476    0.4943    0.6602
    0.5203    0.3955    0.5274    0.2994    0.2422    0.4497
    0.3003    1.2505    0.6264    0.6104    0.0731    0.9278
    1.0000    1.0000    1.0000    1.0000    1.0000    1.0000
    1.1268    0.1269    1.3431    0.6588    0.1034    0.3995];
 

P=b(:,[1,2,3,5]);
T=[1 0 1 2];
x=[1 1 1 1];
%繪出訓(xùn)練樣品數(shù)據(jù)
figure;
plot(T,x,'r*')
%設(shè)置網(wǎng)絡(luò)隱單元的神經(jīng)元數(shù)(可
 n=6;
%建立BP網(wǎng)絡(luò)
net=newff(minmax(P),[n,1],{'tansig','logsig'},'trainlm');
%訓(xùn)練建立的網(wǎng)絡(luò)
 net.trainParam.show=50;  
 net.trainParam.lr=0.05;
 net.trainParam.mc=0.9;
 net.trainParam.epochs=2000;
 net.trainParam.goal=0.001;
%調(diào)用TRAINGDM算法訓(xùn)練BP網(wǎng)絡(luò)
 net=train(net,P,T);
%對BP網(wǎng)絡(luò)進(jìn)行仿真,并計算誤差
 A=sim(net,P)
 E=A-T;
 M=sse(E)
 N=mse(E)
%對建立的網(wǎng)絡(luò)進(jìn)行仿真,對未知的樣品進(jìn)行分類
p1=b(:,[4,6]);
y=sim(net,p1)
%繪出仿真樣品數(shù)據(jù)
z=[1 1];
figure;
plot(T,x,'r*')
hold on
plot(y,z,'bp')